In 2013, the California Independent System Operator (CAISO) introduced the iconic “duck curve” chart depicting the expected effects of variable renewable resources – primarily utility-scale solar – on grid conditions. Since then, the conditions predicted have come to pass: meaningful midday overproduction and late-day ramps. Yet, when measured as a percentage of net generation or total installed capacity, the penetration of solar in the U.S. remains very small, even in California.
The question, then, is how can such a small set of resources have such a large operational impact? Answering this question requires a clear understanding of the dynamic relationships between three operating variables in an electric system: annual minimum midday load, peak solar production, and baseload or must-run generation. When taken together, these variables can be used to characterize a stress-case operating scenario and assess the oversupply risks an electric system faces when solar net load (i.e., system load minus solar generation) drops below must-run baseload generation.
This article builds from these dynamic relationships to derive new metrics that estimate the impacts and risks of solar in major U.S. electricity markets as of the end of 2015, the most recent year of data available. More specifically, the analysis examines select North American Electric Reliability Corp. (NERC) regions, as well as regional transmission operators (RTOs) and independent system operators (ISOs). The findings reveal the potential for the electric system to accommodate additional solar, though it is important to note that all results are contingent upon simplifying assumptions made involving an electric system’s operating variables. Changing these assumptions, leveraging advanced capabilities of utility-scale solar, or integrating storage with solar could yield different results.
Overall, the new metrics are valuable because they account for baseload generation considerations while signaling how close a region may be to experiencing oversupply risks from solar generation. Further, the new metrics provide solar stakeholders with a simple heuristic to assess oversupply risks across multiple markets.
Operational impact of solar
Assessing the operational impact of solar requires acknowledging that its effect on an electric system is not well described by simple measures of solar as a percentage of annual generation or total capacity. An increase in total generation capacity will not mitigate the impact of solar on baseload generation needed to meet load. In fact, what matters when discussing variable resources is not what capacity is available, but when that capacity is generating electricity in relation to the load at that time.
Consequently, any operational analysis of solar penetration should consider its effect specifically during midday hours (i.e., 8 a.m. to 8 p.m.). In particular, such an analysis should consider the annual minimum midday system load to assess the maximum operational impact of solar production. An electric system’s baseload or must-run capacity can then be considered to determine whether such a stress-case operating scenario actually poses oversupply risks.
Thus, with just three core variables, we can begin making sense of solar’s operational impact:
A. Minimum Midday Load: The lowest daytime load during a given year. The minimum midday load will often occur on a weekend during a shoulder season (i.e., spring or fall), when heating and cooling demands are lower.
B. Solar Generation: Maximum potential production from existing solar capacity. To ensure a stress-case operating scenario, existing solar capacity should be assumed to be operating at full production.
C. Baseload Generation at Minimum Run: Baseload generation capacity is the thermal capacity expected to run for the majority of the year in a given electric system. The figure can be adjusted by minimum run estimates specific to each thermal resource. Minimum run capacity reflects the ability of a generation resource to ramp down output during low load periods.
Figure 1 shows each of these variables in an illustrative minimum midday load scenario. In addition, the figure highlights two derivative calculations: an electric system’s oversupply cushion and its solar headroom. The oversupply cushion is an estimate of non-baseload generation that can operate during the minimum midday load. This load is typically served by imports or other generation assets, such as renewables or peaking units. The solar headroom calculation represents the amount of new solar capacity that can be added to the electric system before grid operators must contend with oversupply risks.
Such oversupply risks arise when solar generation pushes net load (i.e., the green line) below baseload generation at minimum run (i.e., the orange line). In these instances, the combination of generation from solar and baseload assets exceeds the electric system’s total demand. Consequently, grid operators must pursue mitigation strategies such as curtailment, storage or export.
In light of these observations, two new metrics can be derived, which better account for the operational impact of solar during a stress-case scenario:
Solar Capacity as a Percentage of Minimum Midday Load: This metric reflects the relationship between peak solar production and annual minimum midday load. Using the labels in Figure 1, the metric is calculated in the following manner: Solar Capacity as a Percentage of Minimum Midday Load = B ÷ A
Solar Protection Factor (SPF): This metric signals how close an electric system may be to experiencing oversupply risks from solar generation. A value greater than one provides directional evidence of oversupply risks during the minimum midday load. The metric does not speak to operational impacts associated with ramping. Using the labels in Figure 1, the metric is calculated in the following manner: SPF = B ÷ (A – C)
Compared to traditional solar penetration metrics, these alternatives result in much higher penetration levels of solar across all major electricity markets in the U.S., thus helping to explain the impacts of solar we are witnessing today (see Figure 2). This trend is most evident in CAISO, where 9% of net generation translated into 40% of its minimum midday load and 55% of the ISO’s oversupply cushion (i.e., SPF), thus demonstrating the leverage effect. Also noteworthy is SERC Reliability Corp., whose SPF was more than 23 times greater than its traditional penetration metrics due to the NERC region’s comparatively small oversupply cushion resulting from its unique baseload composition.
It is important to note that this analysis focuses on the worst days of the year, from an operations perspective, to illustrate the point that solar generation’s impact is felt at specific times while not being an issue in annual averages. What’s more, this analysis combines control areas within a given NERC region or RTO/ISO, thus tempering the more dramatic impacts from solar in smaller balancing authorities. A more granular analysis would highlight these impacts and help to explain recent baseload retirements, such as the announced closing of the Diablo Canyon nuclear power plant in CAISO.
Solar oversupply risks
Building from these observations and metrics, we assessed the oversupply risks from solar in multiple electricity markets across the U.S. The analysis is indicative, not definitive. To perform it, we identified annual minimum midday load, installed solar capacity, and baseload generation as of the end of 2015 for select NERC regions and RTO/ISOs. Baseload generation was assumed to be thermal assets with a three-year average capacity factor greater than or equal to 65%. However, we did not consider coal units in SERC estimates to reflect assumptions in stakeholder comments filed in the Federal Energy Regulatory Commission’s recent technical conference on the Public Utility Regulatory Policies Act of 1978. Minimum run was assumed to be 100% for nuclear resources, 80% for coal resources and 60% for combined-cycle natural gas resources.
The results show the CAISO market with the greatest solar penetration during minimum midday load (see Figure 3). Further, the analysis shows CAISO having additional solar headroom (i.e., ability to accommodate additional solar generation) due to the relatively small amount of baseload generation at minimum run. By comparison, the ability to accommodate solar is limited in other electricity markets due to the high percentages of baseload capacity at minimum run. This situation is most pronounced in SERC, where a large reserve capacity and more than 25 GW of nuclear capacity exist. A similar dynamic, proportionally high baseload generation, also appears in the PJM Interconnection.
Our analysis further indicates that in 2015, the Midcontinent Independent System Operator (MISO) could have potentially accommodated more than 20 GW of solar capacity before needing to address oversupply risks from solar generation (see Figure 4). This is reflective of the ISO’s small existing solar capacity at the time as compared to the minimum midday load of more than 58 GW. Finally, it is noteworthy that CAISO’s solar headroom was smaller than all markets except ISO New England (ISO-NE) in 2015. With this insight, it should not be a surprise that CAISO has since begun to increase renewable curtailments in 2017, following the installation of additional solar capacity and a marked increase in hydropower production following drought-busting rainfall.
It should be emphasized that the aforementioned results represent estimates based on simplified assumptions. For example, they do not account for the operational impacts of other variable generation such as wind or factors such as inter-regional trading. It is also worth noting that changes to an electric system’s baseload composition or annual minimum midday load could significantly alter outcomes. For example, if baseload generation capacity increases, solar headroom would shrink regardless of new solar additions. Conversely, if baseload generation capacity decreases due to retirements, then solar headroom could grow. In addition, the analysis does not consider advanced power controls or the integration of energy storage, both of which could allow utility-scale solar to support electric system stability and reliability.
Despite these limitations, the value of this framework and the new metrics, such as the SPF, resides in their ability to quickly approximate oversupply risks using stress-case operating scenarios while accounting for baseload generation. Further, the tools provide grid operators and solar stakeholders with the ability to assess oversupply risks across multiple electricity markets. The new penetration metrics are similar to hosting capacity assessments conducted at the distribution/circuit level and can be used as a milestone signaling the need for focused studies involving more detailed analysis.
In summary, our results suggest there remains room to accommodate solar growth, though electric systems with limited solar headroom or especially large baseload generation capacity (e.g., those with significant nuclear assets) are likely to experience solar oversupply risks well before their peers. A separate analysis would be required to address ramping issues associated with the duck curve’s neck.
All three authors work at energy consulting firm ScottMadden Inc., where Chris Vlahoplus is a partner and cleantech and sustainability practice leader, Paul Quinlan is cleantech manager, and Chris Becker is a research analyst.
